Johns Hopkins University
At A Glance
Name: Jinong Li
Position: Assistant professor of pathology, Johns Hopkins University, since 2003; Research associate and instructor of pathology since 2001.
Background: Postdoc in Ellen Pizer's laboratory, Johns Hopkins University, 1998-2000.
PhD in molecular biology, Umea University in Sweden, 1997.
BS in biochemistry, Beijing University, 1988.
Ciphergen Biosystems said last week that it is developing breast cancer diagnostics, based on a "series of discoveries" that have been made using its SELDI technology. ProteoMonitor talked to Jinong Li, a researcher who made some of those discoveries, to find out about her background and her breast cancer work.
What is your research background, and when did you get into finding biomarkers for breast cancer?
I did my PhD in Sweden. I grew up in China, and after graduation, I went to Sweden to pursue my PhD. I got my PhD from Umea University. I was trained by Glenn Bjork. I was trained as a molecular biologist.
I did my postdoctoral fellowship at Johns Hopkins in the department of pathology, in the laboratory of Ellen Pizer. At that time I was working with her on tumor markers. I studied fatty acid synthase. That's a tumor marker. It's elevated in several different cancers, including breast cancer. So that's how I got into the tumor marker field.
I did two years of post-doctoral fellowship in that laboratory, and then I got my faculty position here in Johns Hopkins under Dr. Dan Chan in clinical chemistry. Actually, when I got this job, I helped Dr. Chan initiate the whole [biomarker] facility, because we just had this center established, but it was basically empty at the time.
Before I came, Dr. Chan started a collaboration with Ciphergen. They sponsored instruments and part of our research. Then we started doing stuff on different types of cancers, but I had my focus on breast cancer from the very beginning.
What projects did you work on in the new biomarker facility?
I think breast cancer was one of the first studies that we actually got good results on. That led to a publication in Clinical Chemistry in 2002. I think it was a pretty important publication from our group, because in that article, we reported about our system — how we analyze serum, and how we do all kinds of things. We identified three serum biomarkers for breast cancer.
What were some of the challenges of getting the biomarker facility started?
I think it was tricky. From my point of view, I didn't have that high-throughput proteomics background. I do have a proteomics background, because I did a lot of 2D gels when I was a graduate student and a postdoc. I was familiar with that, but to do proteomics in a high-throughput manner using those upfront technologies like SELDI was really a new field for me. We needed to learn how to use instruments. We needed to do high-throughput analysis, so we needed to learn new knowledge about bioinformatics tools, which I had never done before.
I joined the center in 2000. We were one of the first groups at that time to start using SELDI.
It was challenging at first to get the protocol to work. We did get a protocol from Ciphergen, but we did a lot of modification on it — that's the wet lab part. In terms of the dry lab part, [a challenge was that] we had never seen that much data before. Once you analyze your sample, you generate so much data, and that's the second problem we encountered. We had to learn how to analyze those data.
What have you followed up upon since you published that three-biomarker breast cancer paper in Clinical Chemistry?
We did a validation study of the first paper. That study was just recently published in 2005 in Clinical Chemistry. That's a follow-up on the 2002 study. And we also did another study on breast fluid.
For the new serum validation study, we used 176 samples. We were able to validate two out of the three markers. The trend was the same — they were elevated in cancer, and low in controls. The expression pattern was validated.
In the first study, we didn't identify the proteins. We reported three peaks at different molecular weights, and we reported their performance — like were they elevated in cancer, or lower in cancer? And we reported how sensitive and specific they were in segregating cancer versus non-cancer.
In the second study, we did identify what proteins the peaks were, and we were able to validate the expression pattern for two out of the three peaks by using an independent cohort. It's really important to do independent validation. You can't use the same sample from the same hospital [as for the first study].
What about the third biomarker?
The third biomarker we were not sure about. It's not that we think it's not a marker. We think the stability of the marker is low. So right now it's a question mark with that third marker.
We are still following up with those markers. Right now we're doing a multi-center study still on serum. We're doing fractionated serum in order to see more protein species. This allows us not only to validate those previous markers, but also to find new markers.
Have you always been working on breast cancer since you joined Hopkins?
Most of my work has been on breast cancer, but I also did a study on prostate cancer that was published in the Journal of Urology a few years ago. It was a similar study with serum biomarkers.
Are you in the process of validating those markers as well?
I have so much to do on breast cancer, so I'm trying to focus more on breast cancer right now.
Why did you decide to use SELDI in your biomarker studies, over other proteomic techniques?
Well, I'd done a lot of gels before. Gels are not high throughput. It's really difficult to analyze gels in a timely fashion. In a clinical setting, if you want to find a marker that can survive a population study, you need to test more samples. You can't just do a few patients and tell people, 'OK, this is a marker. It works in these five patients.'
It's very common that we'll analyze 300 to 400 patient specimens. With gels, that's impossible. You can not analyze that many. I think we used SELDI mainly because of the throughput.
Did you consider using other high-throughput proteomic technologies?
SELDI is one of the high-throughput technologies that you can use. There are other methods that can also offer the throughput that you want. It really depends on what you are looking at.
I think the advantage of SELDI is that it allows you to see intact proteins. What we do is, we run serum samples or fluid samples, and we don't digest samples before we run [them] on protein chips. You get to see the intact proteins or peptides as they are in biological fluid — you don't chop them up.
For some of the other methods that people use, they will chop [the proteins] up first, and then analyze. There are certain advantages and disadvantages to SELDI, but for me, seeing intact peptides is important, and it kind of reduces the complexity of the fluid that you're looking at. Because once you chop a protein up, it becomes five, ten peptides.
Aside from trying to find breast cancer markers in serum, what other studies are you working on?
I'm still working right now on the breast fluid study. I'm fascinated by that, because if you think about it, the serum, where you have the circulation of your whole body, is not as site-specific as the breast fluid. The breast fluid you're taking is directly from breast, so if you want to study fluid, that's probably as close as you can get for breast cancer.
We just published a paper recently in Clinical Chemistry Research. That study was a pilot study with a limited number of samples. Right now we're doing a larger study with more samples.
In that study, we presented some potential biomarkers. One we have identified as human defensin.
How close do you think you are to developing a diagnostic test for breast cancer?
I think that probably will take a long time. All these studies are still small. You need to test these markers really carefully before you do any application. You need to test them on larger cohorts.
With the small cohorts, how well did the markers perform in terms of sensitivity and specificity?
I found them pretty specific, especially for that marker that I presented in Clinical Chemistry Research. But because the number of samples is small, it's really hard to evaluate the sensitivity and specificity. In the real world, you can only evaluate sensitivity and specificity when the cohort is large enough. Otherwise, it's not representative.
Do you have any comment on Ciphergen's recent establishment of a program for breast cancer diagnostics?
I don't know the details about that program, but I assume that they would probably collaborate with as many breast cancer sites as possible and try to come up with a panel of markers.
I think in the near future, we should validate these breast cancer markers using independent cohorts. The first step would be that we need to know what they are. That we have done already for some of the markers. And then, we need to validate them with an independent method. For example, in my publications, the general approach is we first identified those peaks, and then we identified the proteins that correlate to those peaks, and then we validate using independent methods, for example ELISA.
After we validate their performance, then we can really work on combining them together, finding a panel, and stuff.
You validated two of the breast cancer markers. Is that enough to form a panel?
No, I don't think so. We validated two out of the three, but from our results, I don't think the performance is good enough. We do need additional markers to improve the performance. And that's why we did fractionation — so we could see more proteins in serum. So from among that pool, we can find more markers.
I think one direction we need to go in is once we find a potential marker, we can also evaluate those markers in breast cancer tissue, so you know the origin of the marker.
What I think is that there are two types of markers. One type is the markers that are released directly from the breast tumor. The second type is something that is stimulated by breast cancer, but not directly from the tissue. It's the host response towards that tumor.
If you want to study the question of why [the markers] are high in serum, or why they're high in fluid, you need to go back to basic research and see how they work.
Who funds your research?
My research is funded by the Susan Love MD breast cancer foundation, and the Susan G Komen Breast Cancer Foundation, and also the breast cancer SPORE (Specialized Program of Research Excellency). And also Avon. And Ciphergen.
In terms of government funds, the center receives government funds, but I'm not the PI of it. Dr. Dan Chan is the PI of it.